This artifact is only runnable via vLLM's stock compressed-tensors support.

SOTA mixed precision quantization technique (https://github.com/RobTand/prismaquant).

Some layers are NVFP4 - so Blackwell only compatible.

Downloads last month
58
Safetensors
Model size
21B params
Tensor type
F32
·
BF16
·
F8_E4M3
·
U8
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for IstroSec/Qwen3.6-27B-uncensored-heretic-v2-Native-MTP-Preserved-PRISMAQUANT-PRISMASCOUT-6bit